UC Davis UC Davis Previously Published Works

Home , Heptanal, Octanal

Title Chemical and Sensory Characterization of Oxidative Changes in Roasted Almonds Undergoing Accelerated Shelf Life.

Permalink https://escholarship.org/uc/item/61g6r83b

Journal Journal of agricultural and food chemistry, 65(12)

ISSN 0021-8561

Authors Franklin, Lillian M Chapman, Dawn M King, Ellena S et al.

Publication Date 2017-03-21

DOI 10.1021/acs.jafc.6b05357

Peer reviewed

eScholarship.org Powered by the California Digital Library University of California Article

pubs.acs.org/JAFC

Chemical and Sensory Characterization of Oxidative Changes in Roasted Almonds Undergoing Accelerated Shelf Life † ‡ ‡ † § Lillian M. Franklin, Dawn M. Chapman, Ellena S. King, Mallory Mau, Guangwei Huang, † and Alyson E. Mitchell*, † Department of Food Science and Technology, University of California, Davis, One Shields Avenue, Davis, California 95616, United States ‡ Covance Food Solutions, 365 North Canyons Parkway, Suite 101, Livermore, California 94551, United States § Almond Board of California, Suite 1500, 1150 Ninth Street, Modesto, California 95354, United States

*S Supporting Information

ABSTRACT: In almonds, there is no standard method for detecting oxidative changes and little data correlating consumer perception with chemical markers of rancidity. To address this, we measured peroxide values (PV), free fatty acid values (FFAs), conjugated dienes, tocopherols, headspace volatiles, and consumer hedonic response in light roasted (LR) and dark roasted (DR) almonds stored under conditions that promote rancidity development over 12 months. Results demonstrate that, although rancidity develops at different rates in LR and DR almonds, consumer liking was not significantly different between LR and DR almonds. Average hedonic ratings of almonds were found to fall below a designated acceptable score of 5 (“neither like nor dislike”) by 6 months of storage. This did not correspond with recommended industry rejection standard of PV < 5 mequiv peroxide/kg oil and FFA < 1.5% oleic. FFAs remain well below <1.5% oleic during storage, indicating that FFAs are not a good marker of rancidity in roasted almonds stored in low humidity environments. Regression of consumer liking to concentration of rancidity indicators revealed that selected headspace volatiles, including heptanal, octanal, nonanal, 2-octenal, 2-heptanone, 2- pentylfuran, hexanal, and pentanal, had a better correlation with liking than did nonvolatile indicators. KEYWORDS: almond, Prunus dulcis, HS-SPME GC/MS, peroxide value, hedonic analysis

■ INTRODUCTION triglycerides (i.e., oxidative rancidity).4,5 Both processes Sweet almonds (Prunus dulcis) are considered an excellent produce a range of primary lipid oxidation products (e.g., free source of dietary protein, fiber, and micronutrients such as fatty acids, lipid hydroperoxides, and conjugated dienes and vitamin E.1,2 Almonds are also high in unsaturated fatty acids trienes) and secondary lipid breakdown products (e.g., − aldehydes, ketones, alcohols, hydrocarbons, oxiranes, and (44 61% fat by weight) and low in moisture (less than 10% 4,12 moisture w/w) and are subject to lipid oxidation and the lactones). Primary oxidation assessment methods often − development of rancidity during storage.1 3 These changes can include measuring PV and conjugated dienes (CD). Peroxide − result in objectionable flavor and loss of nutritive quality.3 6 value measures the concentration of lipid peroxide, while CD assess the degree of double bond rearrangement co-occurring The two most abundant fatty acids in almonds are oleic acid 13 (18:1, 62−80%) and linoleic acid (18:2, 10−18%).1 Unshelled with the peroxidation of linoleic acid. Measuring FFAs in raw almonds have a shelf life of approximately 12 months due conjunction with lipid peroxidation indicates the degree of to relatively high concentrations of naturally occurring hydrolytic rancidity. PV is the most common method used by tocopherols (∼24 mg/100 g).7 Shelling and processing the industry to estimate oxidation in almonds, while methods such as FFA, CD, and tocopherol quantitation are less almonds exposes nutmeats to light, heat, moisture, and oxygen, 3,7,11,14,15 which can initiate lipid oxidation and shorten shelf life.5 commonly employed. Although lipid oxidation and the development of rancidity is a Because lipid hydroperoxides are themselves imperceptible to persistent problem for processors, there is no completely humans, measuring peroxidation may not directly correlate with fl 12 objective chemical method for determining the onset of avor changes in nuts. Quantitation of headspace volatiles rancidity. To date, the industry relies on indirect measures associated with rancidity (e.g., hexanal) has been used in many foods to better approximate development of rancid flavor and (i.e., peroxide value [PV] and free fatty acids [FFAs]) to 3,5,6,16,17 estimate lipid oxidation in almonds. Currently, there is no aroma. However, human perception of volatiles can ff uniform or standard method for detecting oxidative changes in di er based on unique product characteristics such as surface almonds and, more importantly, there is little data correlating area, moisture content, serving temperature, composition, and consumer perception with these chemical markers of rancid- − ity.3,8 11 Received: November 29, 2016 Rancidity development can occur through hydrothermal or Revised: March 6, 2017 enzymatic hydrolysis of triglycerides (i.e., hydrolytic rancidity) Accepted: March 11, 2017 and/or through the direct oxidation of fatty acids and Published: March 11, 2017

© XXXX American Chemical Society A DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article competing volatiles in the product headspace.18 Additionally, %), 1-H-pyrrole (98%), 1-octanol (99+%), butanoic acid (99+%), 3- sensory testing may not indicate whether the product is methylbutanoic acid (98+%), and nonanoic acid (99+%) were considered acceptable by the end-user, as trained sensory obtained from Sigma-Aldrich (St. Louis, MO, United States). judges should not answer affective questions.18 It is therefore Standards of hexanoic acid (98%), 1-butanol (99%), and ethyl 2- prudent to perform hedonic analysis of food products to (methylthio)-acetate (95%) were obtained from Acros Organics (Thermo Fisher Scientific Inc., Waltham, MA, United States). All benchmark chemical measures with changes in consumer 6 other standards, solvents, and reagents were purchased from either acceptance. Fisher Scientific Company (Fairlawn, NJ, United States) or Sigma- A number of studies have assessed consumer acceptance in Aldrich Chemical Co. These included the following: HPLC/ 3,9−11,15 relation to chemical measures of rancidity in almonds. spectrophotometric grade solvents ethanol, methanol, methyl tert- Most3,10,11,15 have focused on the effect of packaging type on butyl ether, glacial acetic acid, chloroform, and 2,2,4-trimethylpentane; consumer acceptance (and chemical markers of oxidation) analytical grade sodium hydroxide, hydrochloric acid (ACS certified during storage. For example, Mexis and Kontominas measured 36.5−38% w/w), potassium iodide (99.9%), and sodium thiosulfate (99%); and analytical standards of (±)-α-tocopherol, (+)-δ-tocopher- PV, hexanal, color, fatty acid methyl esters (FAME), and γ − consumer acceptance to study the effects of active packaging, ol, (+)- -tocopherol, volatile compounds (95 99%), and conjugated linoleic acid (99%). Exceptions included 2-(ethylthio)-ethanol (96%) nitrogen flushing, and packaging material on the oxidation of 3 (Alfa Aesar, Ward Hill, MA, United States) and 3-hydroxybutan-2-one whole, unpeeled raw almonds. Raisi et al. evaluated the impact (Supelco, Bellefonte, PA, United States). of packaging on PV, conjugated trienes, and hedonic ratings in Roasting and Storage of Samples. A 200 kg sample of dehulled, ground and whole unpeeled almonds for 10 months of storage raw, Nonpareil almond kernels with skin (from the 2014 harvest year) 15 under either vacuum, 95% CO2, or ambient atmosphere. was obtained from Blue Diamond Growers (Sacramento, CA, United Senesi et al. measured the effect of packaging on oxygen and States). Almonds were dry roasted in a BCO-E1 electric convection light permeability, storage atmosphere, and temperature on oven (Bakers Pride, Allen, TX, United States). Kernels were roasted under two different conditions: 115 ± 6 °C for 60 min and 152 ± 6 °C several aspects of oxidation and sensory quality of peeled whole “ ” “ ” raw almonds over 546 days of storage.11 Though a number of for 15 min to produce a light and dark degree of roast, respectively. studies have evaluated the relationship between packaging type, Almonds were divided into 480 g samples and placed in a controlled atmosphere chamber (KMF 240 Constant Climate Chamber by changes in oxidation indicators, and consumer acceptance, no Binder Inc. Bohemia, NY, United States). Samples were stored at 15 ± study directly compares oxidation indictors to each other in 1% relative humidity (RH) and 39 ± 1 °C for intervals of 1−12 their ability to correlate with consumer acceptance of almonds. months. Eight samples each of LR and DR almonds were randomly Additionally, there exists little published literature on the withdrawn from the chamber every month, mixed thoroughly, and volatile profile of roasted almonds during accelerated storage or repackaged into polyethylene vacuum sealed packages which were the relationship between volatile profile, rancidity indicators, then stored at −80 °C (Revco Inc. Trumbull, CT, United States) until and consumer acceptance.19 analyzed. Sample Preparation for Non-Sensory Analysis. Sample Herein, we assess common indicators of lipid oxidation in 20 almonds, including PVs, FFAs, and CD, in conjunction with preparation was adapted from the method of Zhang et al. A 300 g fi α sample of LR and DR almonds were thawed and analyzed within 1 headspace volatile pro les, -tocopherol concentration, and week of removing almonds from controlled storage. An approximate hedonic analysis of roasted almonds to evaluate how volatile 50 g aliquot was removed and ground for nine 1 s pulses (Waring and nonvolatile oxidation indicators correlate with consumer Laboratory Equipment, Torrington, CT, United States). The resulting acceptance in roasted almonds. Data from this study will allow powder was sieved through a 20 mesh Tyler standard screen (W.S. processors to understand how well chemical markers of Tyler Industrial Group, Mentor, OH, United States). The sieved rancidity correlate with consumer liking during rancidity powder was weighed into 20 mL glass headspace vials (Restek development in roasted almonds. Identifying which indicators Corporation, Bellefonte, PA, United States) to give 5.00 ± 0.02 g optimally correlate with consumer liking will ensure processors aliquots, and vials were capped. Oil was extracted from the remaining do not mischaracterize and discard acceptable samples or fail to almond powder with an oil press (Carver, Inc., Wabash, IN, United identify deteriorated samples. States). The resulting oil was collected and transferred to a 40 mL amber borosilicate glass sample vial with a polytetrafluoroethylene (PTFE) lined, rubber faced cap (Fisher Scientific, Pittsburgh, PA, ■ MATERIALS AND METHODS United States), flushed with nitrogen, and stored at −80 °C until Chemicals and Reagents. Stable isotope standards of octanal-d16, analysis. 2-methylpyrazine-d6, and n-hexyl-d13 alcohol, representing three main Peroxide Value, Free Fatty Acids, and Conjugated Dienes. categories of identified compounds (aldehydes, pyrazines, and Peroxide value of each sample was determined according to the AOCS alcohols), were purchased from C/D/N Isotopes Inc. (Pointe-Claire, Official Method Cd 8-53;21 the FFA was determined according to the QC, Canada). Authentic standards of 2-methylpropanal (97+%), AOCS method Cd 3d-63,22 and CD was determined according to butanal (97+%), 3-methylbutanal (97+%), hexanal (99%), heptanal AOCS method Ti la-64.13 (95%), (E)-2-hexanl (98%), octanal (99%), 1-chloro-2-propanol Headspace Solid-Phase Microextraction (HS-SPME) GC/MS (70%), 2,5-dimethylpyrazine (98%), nonanal (95%), furfural (98+ Detection of Headspace Volatiles. Headspace analysis was adapted %), 2-acetylpyrrole (99%), 2-furanmethanol (99%), methyl acetate from the method of Xiao et al.23 Almonds were ground and sieved (99.9%), 2-pentanone (98+%), pentanal (99%), dimethyl disulfide with a 20 Tyler sieve (W.S. Tyler), 5 ± 0.02 g of which was weighed (99+%), 2-methyl-1-propanol (99+%), 2-heptanone (98%), methyl into each 20 mL glass headspace vials (Restek Corporation). Vials hexanoate (99.8+%), 2-nonanone (99.5+%), decanal (95+%), 4- were capped and crimped immediately with magnetic caps containing hydroxy-2,5-dimethylfuran-3-one (99+%), 1-nonanol (98+%), hepta- 3 mm PTFE-lined silicone septa (Supelco Co., Bellefonte, PA, United noic acid (99+%), α-pinene (98%), and octanoic acid (99+%) were States) and allowed to equilibrate for at least 4 h. This equilibration obtained from Aldrich Chemical Co., Inc. (Milwaukee, WI, United period was found to produce the least relative standard deviation States). Authentic standards of 2-butylfuran (98%), 2,3-butanediol among native headspace compounds (see Supporting Information). (97+%), 2-methylpyrazine (99%), (E)-2-octenal (94%), acetic acid All samples were run in triplicate. (99+%), benzaldehyde (99+%), 2-methylbutanal (95+%), 3-methyl-1- To account for variations in fiber and instrument sensitivity, stable butanol (98+%), 1-pentanol (99+%), 2-octanone (99+%), 1-heptanol isotope external standards were run on each day of analysis. Standards (99+%), 1-octen-3-ol (98+%), (E)-2-nonenal (95+%), 1-hexanol (99+ were run externally (in devolatized almond matrix) rather than

B DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article internally to avoid interference of internal standards with the natural Mass spectrometric detection by electron impact ionization was equilibrium between the SPME fiber matrix, headspace volatiles, and performed by an Agilent 5975C inert XL EI/CI MSD (Agilent volatiles in the sample matrix during extraction. In this way, the Technologies) with a source temperature of 230 °C and quadrupole potential competition of internal standards with native compounds for temperature of 150 °C. Volatile profiling was performed using a scan adsorption sites on the SPME fiber was also avoided. Responses of method in the range of 30−300 m/z. Tentative volatile identification external standards were used to normalize the compound peak areas in the resulting total ion chromatogram was performed using the NIST observed in samples on each day of analysis to those of the first day of Mass Spectral Search Program (v. 2.2). Identifications of certain analysis (for sample calculations, see the Supporting Information). compounds were confirmed using retention index calculation and The external standard was prepared as follows: 5 ± 0.02 g of comparison with reference values (Kovat’s Index), and retention time ground, devolatized23 almonds were measured into a 20 mL headspace confirmation with standards when available (Table 3). Integration was vial into which a 200 μL glass vial insert was placed (Thomas performed using Agilent MassHunter Quantitative Analysis software Scientific, Swedesboro, NJ, United States). Devolatized almonds were (v. B.07.00, Agilent Technologies). prepared by holding ground, LR almonds under vacuum (30 mmHg) Tocopherol Concentration by High-Pressure Liquid Chro- at 90 °C for at least 3 days to reduce competing native volatile matography (HPLC)-Fluorescence. This method was adapted from compounds that might influence external standard response, as the Puspitasari-Nienaber et al.24 for the analysis of tocopherol isomers. external standard was intended to indicate fiber and instrument Samples were prepared by diluting 0.200 ± 0.02 g oil in 5 mL of precision. A 0.5 μL glass capillary tube was filled with a 1000 ppm HPLC-grade 1:1 methanol:methyl tert-butyl ether (MTBE) in amber solution of methylpyrazine-d6,hexanol-d13, and octanal-d16 in borosilicate glass vials capped with PTFE-lined caps and agitating for methanol. The filled capillary tube was dropped into the 200 μL vial 20 s. Samples were filtered with 0.2 μm nylon filters (EMD Millipore, insert, and the headspace vial was capped and allowed to equilibrate Billerica, MA, United States) prior to injection. Sample injections of 25 for 2 h, which allowed the entire contents of the capillary tube to μL were separated on a reversed-phase YMC C30 4.6 mm i.d. × 250 volatize and equilibrate with ground almonds. The vial was mm, 5 μm polymeric column at 25 °C. Injection, solvent delivery, and immediately analyzed after this equilibration period. fluorescence detection were accomplished by an Agilent 1200 HPLC Volatiles were quantitated by comparing the extracted ion peak system (Agilent Technologies). The method was optimized for ff areas to those of a relative standard, either methylpyrazine-d6, hexanol- purposes of separating 3 of 4 tocopherol isomers and e ectively 24 d13, or octanal-d16. These were chosen to represent the range of removing long-chain triglycerides and fatty acids from the column. structures and polarities of the most highly concentrated native Solvent A consisted of 100% methanol, while solvent B consisted of headspace compounds. Relative standard curves were prepared to give 100% MTBE. The solvent program proceeded as follows: a 15 min final concentrations (μg/kg ground almond) in the expected linear gradient from 100% solvent A to 65:35 solvent A:B, a 3 min concentration ranges of the native headspace compounds. A 5 μL linear gradient from 65:35 solvent A:B to 100% solvent B, which was capillary tube of methanolic standard mixture was dropped into a held for 2 min, and finally a 3 min gradient to 100% solvent A which preplaced vial insert above 5 g of ground, devolatized roasted almonds was held for 2 min. Fluorescence detection was accomplished at an within a 20 mL headspace vial; the vial was capped, and the system was excitation and emission wavelengths of 293 and 325 nm, respectively. left to equilibrate for 4 h before instrument sampling (to reflect sample Pure standards of α-, δ-, and γ-tocopherol were diluted in methanol to preparation). The relative standard curve was prepared from the a stock concentration of 200 mg/mL. Actual concentration of stock response of hexanol-d13 and octanal-d16 at concentration levels of solutions was determined using UV−vis spectrophotometry at an 2000, 1000, 500, 200, 100, 40, 20, 10, and 5 μg/kg almond and the absorption wavelength of 292 nm for α-tocopherol, and 298 nm for δ- response of methylpyrazine-d6 at 200, 100, 50, 20, 10, 4, 2, 1, and 0.5 and γ-tocopherol, respectively, and extinction coefficients of 75.8, 91.4, μg/kg almond. The limits of detection (LOD) were 0.254, 1.76, and and 87.3 ε1% 1 cm, respectively.25 Working standards of 40, 20, 10, 5, μ 0.407 g/kg almond for methylpyrazine-d6, octanal-d16, and hexanol- 1, 0.5, 0.1, 0.05, and 0.01 ppm were created from stock solution in d13, respectively, while the limits of quantitation (LOQ) were 0.771, serial dilution to create the curve for each standard. Standard curves of 5.33, and 1.23 μg/kg almond, respectively. These values were α- and δ-tocopherol were found to be linear from 0.1 to 40 ppm (r2 = calculated according to eq 1: 0.999), while γ-tocopherol was found to be linear from 0.1 to 10 ppm (r2 = 0.999). According to eq 1, the LODs of α-, δ-, and γ-tocopherol Fσ limit of detection or quantitation = were 0.025, 0.19, and 0.0038 ppm respectively, while the LOQs were slope (1) 0.085, 0.063, and 0.013 ppm, respectively. Consumer Hedonic Analysis. Ninety-nine untrained consumers where F is equal to 3.3 for the LOD or 10 for the LOQ, σ is the between the ages of 14 and 80, who were not pregnant and consumed standard deviation of the y-intercept of the linear regression fitted to almonds at least once a month, were recruited in the city of Davis, the standard curve, and slope is the slope of the linear regression fitted California for hedonic analysis. Consumers were served samples (6−7 to the standard curve. The calculation for determining the relative almonds), identified by three-digit random numbers, at room concentration of analytes (RCAnalyte) is given in the Supporting temperature. Consumers were asked to taste two almonds at a time Information and indicate their liking of samples by marking a nine-point hedonic Sample handling and gas chromatography were performed using an scale.16,18 Participants were given verbal and written instructions, a tray Agilent 7890A gas chromatograph (Agilent Technologies, Santa Clara, of samples, a paper ballot, and sat voluntarily at seats separated by CA, United States) equipped with a CTC Combi/PAL autosampler dividers at which distilled water, an expectoration cup, and unsalted (CTC Analytics AG, Zwingen, Switzerland). Samples were agitated at wheat crackers were provided for palate cleansing. Consumers tasted 500 rpm and pre-equilibrated at 40 °C for 45 min, after which they samples in a random and counterbalanced order to ensure that carry- were extracted with a 1 cm 30/50 μm StableFlex divinylbenzene/ over effects were minimized. carboxen/polydimethylsiloxane fiber at a depth of 29 mm for 45 min Statistics. Results from all analyses were analyzed with a two-way at 250 rpm. After extraction, the fiber was desorbed in a splitless analysis of variance (ANOVA) with interactions and, when injection at 250 °C for 0.9 min, at which time the split vent was appropriate, a two-way multivariate analysis of variance (MANOVA) opened at a 50:1 split for a total injection time of 10 min. The fiber with interactions, testing roast level and sample age as main effects. was then desorbed in a separated helium-flushed needle-heater for 5 Where appropriate, a Bonferroni correction was made to the p-value to min to prevent carryover effects. The headspace volatiles were adjust for multiple iterations of ANOVA. When significant main effects separated using a 30 m × 0.25 mm × 0.25 μm Agilent DB-Wax were found, Tukey’s honestly significant difference posthoc testing was column (Agilent Technologies) with a helium flow rate of 1.2 mL/min performed. A correlation matrix was constructed of all chemical at 35 °C for 1 min then a ramp of 3 °C/min until 65 °C was attained, measures and average hedonic ratings, and their Pearson’s correlation followed by a ramp of 6 °C/min to 180 °C, and finally 30 °C/min to r-value and p-value of correlation were calculated. Statistical analysis 250 °C, which was held for 5 min. was performed primarily using R and R studio software, including base

C DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article statistical analysis packages, as well as SensoMineR, plyr, dplyr, ggplot2, agricolae, and hmisc.26 value 2 R ■ RESULTS AND DISCUSSION Almonds were roasted under conditions to produce LR or DR almonds and stored at 39 ± 1 °C and 15% relative humidity for nd Correlation 12 months. Temperature conditions were chosen to be high enough to accelerate the development of lipid peroxidation and rancidity over the typical shelf life of roasted almonds7,27 but 10 0.001 a 0.540 0.9 ab0.30 bc 0.721 0.625 0.08 d0.03 a 0.489 0.761 11 f 0.814

not so high as to alter the typical progress of rancidity. ± ± ± ± ± Humidity levels were chosen to avoid increasing the moisture ± content of almonds28 and thereby altering the texture, as this might produce an unintended negative bias in consumer perception.29 Rancidity development was evaluated by measuring PV, 0.002 b 0.302 1.0 ab0.27 bc 31.7 8.24 FFAs, CDs, total tocopherols, and volatile organic compounds 0.08 b0.03 a 1.26 0.36 32 def 334 2.1 d ± ± ± ± ± ±

by HS-SPME-GC/MS at 1-month intervals over the 12-month ± storage period. Consumer hedonic testing was performed on samples to correlate chemical markers of lipid oxidation in almonds with consumer preference. Consumers (99) performed a hedonic rating of the LR and DR samples at 0 (control), 2, 4, 6, 8, and 10 months. ANOVA 0.008 bc 0.267 0.10 c 8.61 0.4 ab 33.5 0.18 c0.02 ab 2.70 0.36

fi ff 4 def 363 2.0 d 4.7 demonstrated a signi cant di erence in liking related to the ± ± ± ± ± ± storage time of the sample, but no significant difference was ± s HSD testing. found in liking related to the degree of roasting (i.e., LR or ’ DR). The average hedonic score was highest at time 0 (7.2 ± 1.7 for DR and 7.4 ± 1.4 for LR) and decreased with the time the almonds were stored, reaching a low of 4.2 ± 2.0 and 4.7 ± 0.003 e 0.261 0.32 c 8.10 0.5 ab 31.9 0.02 b0.01 abc 2.14 0.32

2.1 for DR and LR, respectively (Table 1 and 2). Almonds 3 bcd 362 1.9 d 4.9 ± ± ± ± ± ± stored for 12 months were noticeably rancid and were not used ± to perform consumer hedonic rating. Tukey’s HSD posthoc analysis revealed that, on average, consumers had a significant difference in liking between samples aged 0, 2, 4, and 6 months, while there was no significant difference found between accelerated storage time (months) samples aged 6, 8, and 10 months (Table 1 and 2). 0.000 cd 0.242 0.10 abc 7.95 0.3 ab 32.8 0.20 d0.00 bc 2.84 0.31 1 abc 387 1.55 c 4.9 Peroxide value (PV) is commonly used in the industry to ± ± ± ± ± ± indicate almond quality8,14,15 and, in general, almonds are ± considered to not have undergone significant lipid oxidation if values are <5 mequiv peroxide/kg oil.8 Herein, the initial average PV in the DR and LR almonds were 0.61 and 0.58 ±

0.04 mequiv peroxide/kg oil, respectively (Tables 1 and 2). a erent from other values on the same line by Tukey 0.000 f 0.255 0.07 bc 8.68 0.2 ab 34.1 9 0.14 d0.00 cd 1.38 0.28 ff 16 a 406 1.5 b 5.8 These results are in agreement with results of Mexis et al. for ± ± ± ± ± ± PV of fresh roasted almonds. PVs were found to differ ± significantly with relation to storage time as well as between cantly di roast level, while the correlation of PV with consumer liking fi had an R2 of only 0.489 (significant at p < 0.05). DR almonds displayed a maximum PV between 8 and 10 months and 0.000 f 0.222 0.09 ab 8.35 0.5 ab 34.0 0.04 e0.00 d 1.13 0.25 exceeded 5 mequiv peroxide/kg oil by 5 months of storage 4 a 424 1.4 a 6.6 ± ± ± ± ± ± (Table 2, Figure 1A). The PVs began to decline after 10 ± months and reflected expected degradation to secondary oxidation products. In contrast, the LR samples displayed overall PVs lower than those of the DR samples (Tables 1 and 2, Figure 1A), and never exceeded the 5 mequiv industry- recommended limit. Maximum PV in LR almonds was observed between 6 and 8 months (Figure 1A, Table 1). PVs for both types of almond samples had changed significantly by two months of storage.

FFAs, measured as % oleic, in both LR and DR almonds analysis 0 2 4 6 8 10 12 correlation increased over the 12 months of storage (Figure 1B) and were signiﬁcantly diﬀerent between LR and DR almonds. Levels of

2 -tocopherol conc (mg/kg) 34.3 FFAs correlated significantly with liking (R = 0.761) at p < γ Values of Chemical Analysis versus Mean Hedonic Score -tocopherol conc (mg/kg oil) 435 -+ 0.05 and differed significantly from an initial value of 0.20 ± -tocopherol conc (mg/kg) 9.38 δ mean consumer hedonic score 7.4 β α peroxide value (meq peroxide/kg oil)free fatty acid value (%conjugated oleic) dienes (%) 0.57 0.21 0.213 mode of consumer hedonic score 8 7 6 4 4 4 2 Values followed by the same letter were not found signi

0.03% oleic by 2 months in DR samples and from an initial Table 1. Average Values andR Standard Deviations Resulting from Chemical Analyses and Consumer Hedonic Analysis of LR Almonds for Even Storage Times a a

D DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article

value of 0.21 ± 0.00% oleic by 4 months in LR samples (Tables

value 1 and 2). By 12 months of storage, levels of FFAs in DR and

2 ± ± R LR almonds increased to 0.57 0.04 and 0.36 0.03% oleic, respectively, below the industry-recommended maximum value of <1.5% oleic (Tables 1 and 2).8

nd Correlation FFA is a measure of hydrolytic rancidity occurring either by enzymatic or spontaneous hydrolysis of triglycerides.5,14 During almond roasting, naturally occurring lipases and other enzymes that may hydrolyze triglycerides are inactivated.14 The roasting 0.03 i 0.489 0.001 a 0.540 0.04 a 0.761 1.3 fg 0.721 0.20 bc 0.625 process also reduces almond moisture content, making it less 9 h 0.814 ± ± ± ± ± ± available for lipolysis, a condition maintained by the low RH of sample storage conditions. The dependence of FFA values on storage RH was demonstrated by Lin et al., who showed that whole blanched almond samples stored at 37.8 °C and 35% RH remained below 0.3% oleic for 8 months of storage, while 0.18 g 13.00 0.002 c 0.508 samples stored at 65% RH had increased to 0.4% oleic, and 0.04 c 0.57 0.1 g 31.2 0.04 bcd 8.23 6 h 292 2.0 d ± ± ± ± ± samples stored at 85% had increased to over 0.8% oleic in the ± ± same period.27 Additionally, unheated samples showedr increases in FFA greate than those in blanched samples, attributed to greater enzyme activity.27 Enzyme activity and above-average humidity are of little concern if almonds are roasted and packaged under controlled atmosphere, and FFA 0.53 g 16.48 0.003 c 0.464 0.04 c 0.54 0.6 efg 31.0 0.15 cd 8.10 8 fg 277 2.0 d 4.2 ± ± levels may not exceed 1.5% oleic during extended storage ± ± ± ± 27 ± despite increases in PV and other indicators of rancidity 27

s HSD testing. (Tables 1 and 2). Our results and those of Lin at al. indicate ’ that the existing limit of <1.5% FFA is a more useful indicator of rancidity development and possibly compromised quality in raw almonds exposed to atmospheric moisture. 0.23 j 16.07 0.001 e 0.464 0.01 e 0.42 0.3 cdefg 32.0 0.30 d 7.98 Similar to the FFA, relative changes in levels of CD were 8 de 328 2.0 d 4.5 ± ± ± ± ± ﬁ ±

± signi cant over the 12 months of storage (Tables 1 and 2), and levels were significantly different between roast level. CD correlated significantly with liking (R2 = 0.540) at p < 0.01. Initial levels of CDs in DR and LR almonds were 0.22 ± 0.00 accelerated storage time (months) and 0.21 ± 0.00%, respectively. At 12 months of storage, values ± ±

0.002 g 0.362 increased to 0.51 0.00 (DR) and 0.30 0.00 (LR), an 0.07 l 11.36 0.01 g 0.36 0.5 ab 32.3 0.47 bc 7.52 2 bc 358 1.7 c 4.7 ± ± ± ± ± increase of 135 and 41.7%, respectively (Tables 1 and 2). ± ± Significant increases were first observed at 2 months for the DR samples and at 3 months for LR samples (Tables 1 and 2).4 CD has been found to be a good correlate of PV and indicator of lipid oxidation in soybean, sunflower, and other oils high in

a polyunsaturated fatty acids well as frying fats subjected to high 0.001 i 0.286 erent from other values on the same line by Tukey 0.12 m 3.96 0.00 i 0.37 0.8 abcd 34.4 0.22 ab 8.33 13 ﬀ 7 ab 396 1.4 b 5.8 ± heat. ± ± ± ± ± ± Herein, CD levels in LR and DR almonds displayed a pattern of change similar to that of FFA levels but did not correlate as cantly di

ﬁ well with PV (Figure 1C). CDs may not correlate well with PVs in almonds because CDs are developed from polyunsaturated fatty acids, which account for ∼10−20% of fatty acids in 0.001 j 0.238 0.04 n 2.21 0.03 j 0.18 0.1 a 34.0 0.21 a 8.66 1 a 418

1.7 a 6.8 almonds, while fatty acid hydroperoxides measured by PV may ± ± ± ± ± ± ± develop from both mono- and polyunsaturated fatty acids, which together account for ∼90% of the fatty acids in almonds.1 Measures such as PV and CD commonly applied to liquid oil may have less predictable values in complex substrates without a continuous lipid fraction such as whole nuts.630 CD has not been widely employed to indicate rancidity in nuts, and no recommended limit of CD in almonds currently exists. However, the sensitivity of this assay may be limiting for

analysis 0 2routine 4 screening in 6 an industrial setting. 8 10 12 correlation Tocopherols were resolved as three peaks correlating to α-, δ-, and coeluted β-+γ-tocopherol in almond oil. All isomers -tocopherol conc (mg/kg) 35.0 γ were found to vary significantly over storage time, while only α- Values of Chemical Analysis versus Mean Hedonic Score -tocopherol conc (mg/kg oil) 444 -+ -tocopherol conc (mg/kg) 9.05 tocopherol was found to be significantly different between roast peroxide value (meq peroxide/kg oil) 0.61 free fatty acid value (%conjugated oleic) dienes (%) 0.20 0.216 α mode of consumer hedonic score 8 8 7 4 4 4 β δ mean consumer hedonic score 7.2 2 Values followed by the same letter were not found signi

Table 2. Average Values andR Standard Deviations Resulting from Chemical Analyses and Consumer Hedonic Analysis of DR Almonds for Even Storage Times a a levels (Figure 1D). All isomers were found also to correlate

E DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article

Figure 1. Peroxide value (A), free fatty acid value (B), conjugated diene value (C), and α-tocopherol concentration (D) at each month of storage time. Values for LR almonds are depicted with gray, and DR almonds are depicted with black. significantly with consumer liking at p < 0.01. α-Tocopherol Under accelerated storage conditions, heat and atmospheric had the highest R2 value at 0.814, whereas β-+γ- and δ- oxygen promote the peroxidation and decomposition of oleic tocopherol had correlation R2 values of 0.721 and 0.625, and linoleic acid.4,17 Of the volatiles identified, several may respectively. Initial average levels of α-tocopherol in DR and LR result from oleic acid hydroperoxide decomposition, including almonds were 444 ± 1 and 435 ± 4 mg/kg oil, respectively octanal, heptanal, nonanal, octane, 2-decenal, 2-undecenal, 4 (Tables 1 and 2), and these values decreased significantly at 3 decanal, and heptane (Table 3). Cleavage of linoleic months in DR almonds and 5 months in LR almonds. After 12 hydroperoxides can result in production of hexanal, pentanal, months of storage, levels of α-tocopherol decreased by 34.3 and 1-pentanol, 2-pentylfuran, 2-octenal and 3-octen-2-one, 2- 23.2% in DR and LR almonds, respectfully; levels of β-+γ- hexenal, 2,4-decadienal, 2,4-nonadienal, 2-heptenal, and 1- − δ octen-3-ol, all of which were identified in LR and DR headspace tocopherol decreased by 7.72 10.5%, and -tocopherol 4 decreased by 9.08−12.1% (Table 1 and 2). (Table 3). All volatile products of oleic and linoleic acid were A total of 98 volatile compounds were detected in roasted found to increase over the course of storage. almonds over the 12 months of accelerated storage (Table 3). Organic acids such as acetic, pentanoic, and hexanoic acid have been previously identified as possible tertiary products of Fifty compounds were confirmed with authentic standards lipid oxidation.19,31 Acetic, pentanoic, and hexanoic acid were (Table 3). Tentative identities of the remaining 48 compounds among the organic acids identified in almond headspace (Table were made by comparing MS spectra with the NIST 14 Library ’ 3). These and other acids found in almond headspace were (NIST Library Search program) and calculating Kovat s possibly generated by the oxidation of related saturated retention indixes (KI) with literature values of standards 19,23 aldehydes or autoxidation of unsaturated aldehydes such as chromatographed under comparable conditions. 12,31 fi ff 2,4-decadienal. Sample storage time was found to have a signi cant e ect on Application of heat in low moisture food systems can all volatile compounds, whereas degree of roasting was found to instigate complex Maillard reactions which contribute charac- fi be signi cant for all but 25 of the compounds (Table 3). In teristic flavors to many cooked foods, including nuts.32 A general, levels of organic acids, aldehydes, and alcohols (>4 variety of Maillard reaction products were identified in almond carbons in length) increased during storage, whereas alcohols headspace, including several pyrazines, pyrroles, furans, 4- with 4 or fewer carbons in length decreased. Pyrazines generally hydroxy-2,5-dimethylfuran-3-one, methanethiol, dimethyldisul- decreased over the 12 months of storage, as did ketones and fide, 2,3-pentanedione, 1-(acetyloxy)-2-propanone, 2-methyl- esters with less than 6 carbons. Ketones and esters with 6 or propanal, and 2- and 3-methylbutanal (Table 3).19,32 During more carbons increased over the course of 12 months of Maillard reactions, 2-methylpropanal, 2- and 3-methylbutanal, storage (Figures 2A−P). and a variety of alkylpyrazines are generated by Strecker

F DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article Table 3. Compounds Tentatively Identiﬁed and Quantitated in LR and DR Almond Headspace

a b tR standard unknown literature KI quant. compound group volatile compound external standard CAS number unknown KIb KIb (NIST) ionc d organic acid 3-methylbutanoic acid hexanol-d13 503-74-2 23.22 1670 1681 1680 60.1 di acetic acid hexanol-d13 64-19-7 18.20 1427 1427 1429 60.1 d butanoic acid hexanol-d13 107-92-6 22.37 1630 1639 1650 60.1 d heptanoic acid hexanol-d13 111-14-8 28.41 1942 1948 1954 60.1 d hexanoic acid hexanol-d13 106-70-7 26.39 1842 1842 1849 60.1 d nonanoic acid hexanol-d13 112-05-0 31.39 2099 2101 2144 60.1 d octanoic acid hexanol-d13 124-07-2 30.20 2034 2040 2038 60.1 e pentanoic acid hexanol-d13 109-52-4 24.46 1744 1725 60.1 f d low mol. wt. alcohol 3-methyl-1-butanol hexanol-d13 123-51-3 11.31 1212 1212 1209 55.1 e 1,2-propanediol hexanol-d13 627-69-0 21.48 1592 1599 45.1 e 2-hydroxypropyl acetate hexanol-d13 57-55-6 21.08 1571 1579 74.1 d 1-butanol hexanol-d13 71-36-3 9.00 1138 1140 1145 56.1 dg 2,3-butanediol hexanol-d13 513-85-9 20.72 1552 1553 1542 45.1 e 2-chloro-1-propanol hexanol-d13 78-89-7 16.17 1364 1376 57.1 di 1-chloro-2-propanol hexanol-d13 127-00-4 14.68 1316 1317 1314 45.1 di 2-methyl-1-propanol hexanol-d13 78-83-1 7.25 1093 1086 1092 84.1 f d high mol. wt. alcohol 1-heptanol hexanol-d13 111-70-6 18.56 1441 1442 1467 70.1 d 1-hexanol hexanol-d13 111-27-3 15.97 1356 1357 1355 56.1 d 1-nonanol hexanol-d13 143-08-8 22.95 1667 1668 1661 56.1 d 1-octanol hexanol-d13 111-87-5 20.86 1559 1560 1553 69.1 d 1-octen-3-ol hexanol-d13 3391-86-4 18.43 1436 1434 1430 57.1 d 1-pentanol hexanol-d13 71-41-0 12.89 1259 1261 1255 55.1 di 2-furanmethanol hexanol-d13 98-00-0 22.83 1668 1661 1660 98.1 e 3-heptanol hexanol-d13 589-82-2 14.35 1307 1306 69.1 g di low mol. wt. aldehyde 2-methylbutanal octanal-d16 96-17-3 3.13 902 893 909 57.1 di 2-methylpropanal octanal-d16 78-84-2 2.14 818 821 819 72.1 di 3-methylbutanal octanal-d16 590-86-3 3.19 899 897 925 58.1 d butanal octanal-d16 123-72-8 2.68 858 860 867 72.1 e high mol. wt. (E,E)-2,4-decadienal octanal-d16 25152-84-5 25.69 1808 1807 81.1 g aldehyde e (E,E)-2,4-nonadienal octanal-d16 5910-87-2 23.63 1702 1701 81.1 e (Z)-2-decenal octanal-d16 2497-25-8 22.54 1645 1644 70.1 e (Z)-2-heptenal octanal-d16 57266-86-1 14.86 1322 1319 83.1 d (E)-2-hexenal octanal-d16 6728-26-3 11.36 1212 1213 1204 69.1 d (E)-2-nonenal octanal-d16 18829-56-6 20.24 1526 1527 1530 83.1 d (E)-2-octenal octanal-d16 2548-87-0 17.73 1411 1412 1412 70.1 e (E)-2-undecenal octanal-d16 53448-07-0 24.63 1754 1722 70.1 d benzaldehyde octanal-d16 100-52-7 19.81 1505 1507 1502 105 d decanal octanal-d16 112-31-2 19.44 1487 1488 1484 57.1 d heptanal octanal-d16 111-71-7 10.26 1177 1179 1174 70.1 di hexanal octanal-d16 66-25-1 6.85 1068 1074 1084 57.1 d nonanal octanal-d16 124-19-6 16.90 1385 1386 1380 57.1 d octanal octanal-d16 124-13-0 13.93 1292 1293 1280 84.1 d pentanal octanal-d16 110-62-3 4.21 985 973 984 58.1 di ester methyl acetate octanal-d16 79-20-9 2.27 827 829 828 74.1 d methyl hexanoate octanal-d16 142-62-1 10.41 1189 1184 1184 74.1 h ei low mol. wt. ketone 1-(acetyloxy)-2-propanone octanal-d16 592-20-1 18.50 1442 1469 74 ei 2,3-pentanedione octanal-d16 600-14-6 6.15 1051 1058 100.1 d 2-pentanone octanal-d16 107-87-9 4.21 983 972 981 86.1

3-hydroxybutan-2-one octanal-d16 513-86-0 13.61 1289 1283 1284 45.1 (acetoin)di ei acetone octanal-d16 67-64-1 2.16 822 819 58.1 h e high mol. wt. ketone 2-decanone octanal-d16 693-54-9 19.34 1482 1482 58.1 d 2-heptanone octanal-d16 110-43-0 10.15 1174 1176 1170 58.1 d 2-nonanone octanal-d16 821-55-6 16.78 1382 1382 1387 58.1 d 2-octanone octanal-d16 111-13-7 13.79 1288 1289 1297 58.1 e 3-nonen-2-one octanal-d16 14309-57-0 19.72 1501 1506 125.1 e 3-octen-2-one octanal-d16 18402-82-9 17.20 1395 1390 111.1 e alkane 3-ethyl-2-methyl-1,3-hexadiene octanal-d16 61142-36-7 17.33 1399 nd 67.1 e heptane octanal-d16 142-82-5 1.64 684 700 71.1 e octane octanal-d16 111-65-9 2.07 817 800 85.1 i styrene octanal-d16 100-42-5 12.71 1255 1261 104 i toluene octanal-d16 108-88-3 5.52 1032 1042 91.1

G DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article Table 3. continued

a b tR standard unknown literature KI quant. compound group volatile compound external standard CAS number unknown KIb KIb (NIST) ionc furan 2-propylfurane 2-methylpyrazine- 4229-91-8 5.41 1028 1027 81.1 d6 2-butylfurand 2-methylpyrazine- 4466-24-4 8.42 1126 1122 1123 81.1 d6 2-pentylfurane 2-methylpyrazine- 3777-69-3 12.04 1235 1231 81.1 d6 heterocycle 2-acetylpyridineei 2-methylpyrazine- 1122-62-9 21.59 1598 1597 78.1 d6 2-acetylpyrroled 2-methylpyrazine- 1072-83-9 28.44 1947 1949 1949 94.1 d6 4-hydroxy-2,5-dimethylfuran-3- 2-methylpyrazine- 3658-77-3 29.44 2021 1999 1997 128.1 d one d6 furan-2-carbaldehyde (furfural)d 2-methylpyrazine- 98-01-1 18.48 1436 1438 1455 96 d6 1-H-pyrroledi 2-methylpyrazine- 109-97-7 19.73 1500 1502 1498 67.1 d6 5-methyl-2-octylfuran-3-onee 2-methylpyrazine- 57877-72-2 25.76 1811 nd 98.1 d6 lactone γ-hexalactonee 2-methylpyrazine- 695-06-7 23.55 1696 1698 1703 85.1 d6 δ-hexalactonee 2-methylpyrazine- 823-22-3 25.25 1785 1770 70.1 d6 γ-octalactonee 2-methylpyrazine- 104-50-7 27.51 1901 1901 85.1 d6 butyrolactoneei 2-methylpyrazine- 96-48-0 22.01 1619 1626 86.1 d6 pantolactonee 2-methylpyrazine- 599-04-2 29.39 1998 1998 71.1 d6 oxirane butyl oxiranee 2-methylpyrazine- 1436-34-6 5.51 1031 nd 71.1 d6 methoxymethyl oxiranee 2-methylpyrazine- 930-37-0 15.82 1352 nd 45.1 d6 sulfur-containing dimethyl disulfided 2-methylpyrazine- 624-92-0 6.38 1058 1058 1077 94.1 d6 methanethiole 2-methylpyrazine- 74-93-1 1.58 680 665 692 48.1 d6 1-methylthio-2-propanonee 2-methylpyrazine- 14109-72-9 15.07 1328 1293 104 d6 d ethyl 2-(methylthio)-acetate octanal-d16 4455-13-4 18.16 1424 1425 1450 62.1

4-mercapto-4-methyl-2- hexanol-d13 255391-65-2 20.08 1609 1520 1535 75.1 pentanole terpene 3-carenee 2-methylpyrazine- 29050-33-7 8.75 1133 1135 44 d6 α-pinened 2-methylpyrazine- 80-56-8 5.10 1018 1019 1026 93.1 d6 o-cymenee 2-methylpyrazine- 527-84-4 13.21 1271 1272 119.1 d6 pyrazine 2-ethyl-6-methylpyrazineei 2-methylpyrazine- 13925-03-6 16.64 1378 1382 121.1 d6 2,3-dimethylpyrazineei 2-methylpyrazine- 5910-89-4 15.18 1335 1332 1337 108.1 d6 2,5-dimethylpyrazined 2-methylpyrazine- 123-32-0 14.84 1320 1322 1320 108.1 d6 2,6-dimethylpyrazineei 2-methylpyrazine- 108-50-9 15.03 1324 1328 1325 108.1 d6 2-ethenyl-6-methylpyrazineei 2-methylpyrazine- 13925-09-2 19.30 1480 1488 120.1 d6 2-ethyl-3,5-dimethylpyrazinee 2-methylpyrazine- 13067-27-1 18.58 1441 1443 1444 135.1 d6 2-ethyl-5-methylpyrazinee 2-methylpyrazine- 13360-64-0 17.13 1393 1397 121.1 d6 3-ethyl-2,5-dimethylpyrazinee 2-methylpyrazine- 13360-65-1 18.19 1425 1426 1430 135.1 d6 2-ethylpyrazineei 2-methylpyrazine- 13925-00-3 15.18 1331 1332 1337 107.1 d6 2-methylpyrazined 2-methylpyrazine- 109-08-0 13.04 1264 1266 1267 94.1 d6 pyrazinamideei 2-methylpyrazine- 98-96-4 23.87 1714 1740 80.1 d6

H DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article Table 3. continued

a b tR standard unknown literature KI quant. compound group volatile compound external standard CAS number unknown KIb KIb (NIST) ionc pyrazineei 2-methylpyrazine- 290-37-9 11.12 1204 1206 1204 80.1 d6 2,3,5-trimethylpyrazinee 2-methylpyrazine- 14667-55-1 17.13 1397 1393 1402 122.1 d6 a b ’ tR, retention time. KI, Kovat s retention index based on 30m DB-Wax column; literature values were obtained from NIST Chemistry WebBook, http://webbook.nist.gov/chemistry/Quant. cIon: extracted ion from total ion scan used for quantitation. dCompound identity confirmed with authentic standard. eCompound tentatively identified based on its MS fragmentation pattern and similarity of calculated Kovat’s retention index with values from literature. fLow molecular weight or high molecular weight alcohol, indicating ≤4 carbons in length and >4 carbons in length, respectively. gLow molecular weight or high molecular weight aldehyde, indicating ≤4 carbons in length and >4 carbons in length, respectively. hLow molecular weight or high molecular weight ketone, indicating ≤5 carbons in length and >5 carbons in length, respectively. iCompound not found to be significantly different across roast level at p < 0.05.

Figure 2. Concentration of summed volatiles in DR and LR almonds (μg/kg) over time, grouped by structural similarity. Compound group membership is displayed in Table 3. degradation of leucine, isoleucine, and glycine, while furans and fresh roasted almonds and decreased significantly by two furanones originate from cyclization and dehydration of the months of storage (Table 4). Amadori compound.32 Methanethiol and dimethyldisulfide are Maillard products 2-methylpropanal and 2,3-methylbutanal degradation products of methionine,33 while both 2,3- have low odor thresholds33 (Table 5) and contribute a malty, pentanedione and 1-(acetyloxy)-2-propanone have been nutty aroma to foods, while alkylpyrazines contribute a roasty, shown to result from heating of Maillard compound 2,3- nutty, and earthy aroma to peanuts35 and have been reported in dihydro-3,5-dihydroxy-6-methyl-4(H)-pyran-4-one.34 toasted almonds.19,23,36 2,5-Dimethylpyrazine and 2-methylpyr- The degree of liking of LR and DR almonds changed azine were the pyrazines found in highest abundance in fresh significantly from the control almonds by 2 months of storage. roasted almonds (Table 4), but ethylmethylpyrazines are Volatile compounds for which there was a significant reported to have low sensory thresholds (<1 μg/kg) and may correlation (p < 0.05) and a significant positive or negative also be important to toasted almond aroma. Furfural and 2- change in concentration at 2 months are shown in Table 4.Of furanmethanol were identified in toasted almonds and have these 47 compounds, 36 are positively correlated with sweet/almond and cooked sugar aromas, respectively, but due consumer liking and are primarily Maillard reaction products. to a high sensory threshold (>1 mg/L), furans were reported These compounds were found in highest concentrations in unlikely to contribute to toasted almond aroma.36 1-

I DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article

Table 4. Change in Concentration from 0 to 6 Months of Compounds with Signiﬁcant Correlation (p < 0.05) to Consumer a Liking

concentration in DR almonds concentration in LR almonds Pearson’s r name value 0 months 2 months 6 months 0 months 2 months 6 months positive correlations 1,2-propanediol 0.940 127 ± 4 124 ± 4 37.4 ± 5.2 128 ± 8 75.8 ± 0.3 67.1 ± 1.4 methoxymethyl oxirane 0.939 3.70 ± 0.00 2.27 ± 0.03 0.604 ± 0.020 4.73 ± 0.25 2.80 ± 0.05 0.744 ± 0.007 2-chloro-1-propanol 0.936 0.601 ± 0.003 0.592 ± 0.007 nd 0.706 ± 0.020 0.399 ± 0.009 nd 1-chloro-2-propanol 0.927 52.1 ± 0.1 47.4 ± 0.2 11.7 ± 0.3 60.0 ± 0.3 31.5 ± 0.2 12.3 ± 0.1 2-ethylpyrazine 0.921 5.56 ± 0.08 3.45 ± 0.09 2.11 ± 0.05 5.11 ± 0.15 3.28 ± 0.04 2.28 ± 0.02 2,3-dimethylpyrazine 0.918 3.67 ± 0.04 2.36 ± 0.11 1.37 ± 0.04 3.22 ± 0.05 2.05 ± 0.02 1.54 ± 0.15 2-ethyl-6-methylpyrazine 0.908 5.41 ± 0.10 3.59 ± 0.10 2.13 ± 0.05 4.24 ± 0.17 3.16 ± 0.06 2.35 ± 0.06 toluene 0.895 4.58 ± 0.10 6.39 ± 0.27 5.32 ± 0.60 6.93 ± 0.08 4.14 ± 0.11 3.75 ± 0.19 2,5-dimethylpyrazine 0.886 80.1 ± 1.4 49.2 ± 1.3 28.8 ± 0.7 60.6 ± 1.5 40.3 ± 0.4 29.8 ± 0.7 3-methylbutanal 0.886 92.9 ± 0.8 64.4 ± 0.9 11.2 ± 0.7 114 ± 1 46.4 ± 1.0 8.28 ± 0.14 2-methylbutanal 0.884 225 ± 3 146 ± 2 23.1 ± 1.8 260 ± 5 106 ± 2 18.8 ± 0.2 3-ethyl-2,5-dimethylpyrazine 0.880 9.22 ± 0.22 7.36 ± 0.43 4.79 ± 0.07 7.06 ± 0.27 6.32 ± 0.18 5.01 ± 0.07 2,6-dimethylpyrazine 0.870 13.7 ± 0.3 8.45 ± 0.23 4.25 ± 0.05 8.7 ± 0.45 6.57 ± 0.36 4.78 ± 0.15 1-methylthio-2-propanone 0.869 2.5 ± 0.1 3.15 ± 0.01 nd 4.37 ± 0.10 5.25 ± 0.09 0.752 ± 0.022 2-furanmethanol 0.866 1.85 ± 0.04 1.34 ± 0.06 0.363 ± 0.010 1.61 ± 0.09 0.672 ± 0.008 0.497 ± 0.033 2-ethyl-5-methylpyrazine 0.861 2.11 ± 0.06 1.35 ± 0.035 0.864 ± 0.009 1.51 ± 0.03 1.18 ± 0.02 0.934 ± 0.012 2-methylpyrazine 0.855 44.6 ± 1.8 21.8 ± 0.5 12.6 ± 0.5 35.5 ± 0.6 17.4 ± 0.2 12.1 ± 0.1 pyrazine 0.842 2.49 ± 0.14 1.19 ± 0.05 0.755 ± 0.020 2.41 ± 0.04 0.984 ± 0.038 0.688 ± 0.013 methyl acetate 0.825 15.4 ± 0.1 9.35 ± 0.31 4.31 ± 0.20 20.2 ± 0.1 5.69 ± 0.19 2.52 ± 0.24 1-(acetyloxy)-2-propanone 0.820 266 ± 17 97.8 ± 3.1 41.2 ± 3.4 211 ± 7 71.7 ± 0.3 45.1 ± 0.1 pyrazinamide 0.812 1.32 ± 0.06 0.417 ± 0.011 nd 0.681 ± 0.032 0.317 ± 0.006 nd acetoin 0.807 48.9 ± 1.2 11.3 ± 0.2 2.04 ± 0.07 44.8 ± 1.0 11.4 ± 0.2 2.22 ± 0.06 acetone 0.800 55.6 ± 0.8 33.8 ± 0.5 24.9 ± 2.7 96.4 ± 2.1 34.7 ± 1.8 14.9 ± 0.5 furfural 0.794 14.9 ± 0.4 6.43 ± 0.25 3.95 ± 0.12 15.3 ± 0.7 5.63 ± 0.14 3.78 ± 0.04 2-methylpropanal 0.792 48.1 ± 0.8 16.7 ± 0.6 2.94 ± 0.25 57.4 ± 0.8 11.8 ± 0.6 1.83 ± 0.08 methanethiol 0.783 1.32 ± 0.11 0.318 ± 0.008 0.0841 ± 0.0164 1.84 ± 0.08 0.341 ± 0.028 0.146 ± 0.006 trimethylpyrazine 0.760 13.0 ± 0.3 8.17 ± 0.35 4.93 ± 0.08 7.15 ± 0.31 6.24 ± 0.14 5.02 ± 0.08 2-methyl-1-propanol 0.746 0.44 ± 0.02 2.14 ± 0.01 0.832 ± 0.067 1.51 ± 0.02 2.01 ± 0.05 0.797 ± 0.016 pyrrole 0.727 14.0 ± 0.5 1.65 ± 0.03 0.243 ± 0.006 9.47 ± 0.15 1.62 ± 0.05 0.393 ± 0.027 2,3-pentanedione 0.721 28.6 ± 0.6 2.15 ± 0.08 0.409 ± 0.021 25.7 ± 0.7 1.96 ± 0.05 0.449 ± 0.023 4-mercapto-4-methyl-2- 0.721 5.88 ± 0.15 nd nd 2.99 ± 0.06 nd nd pentanol dimethyl disulfide 0.714 0.857 ± 0.002 2.36 ± 0.103 nd 1.78 ± 0.0322 3.91 ± 0.128 0.325 ± 0.012 2-acetylpyrrole 0.713 0.821 ± 0.0168 0.530 ± 0.019 0.405 ± 0.0176 1.02 ± 0.06 0.759 ± 0.017 0.591 ± 0.016 negative correlation hexanoic acid −0.900 1.05 ± 0.06 14.7 ± 3.0 128 ± 4 1.96 ± 0.32 12.3 ± 0.2 106 ± 3 pentanal −0.902 3.92 ± 0.04 62.8 ± 0.6 241 ± 10 5.5 ± 0.1 41.3 ± 0.8 123 ± 1 hexanal −0.902 58.0 ± 0.3 716 ± 16 2380 ± 40 77.9 ± 1.8 492 ± 9 1480 ± 10 2-pentylfuran −0.909 4.86 ± 0.14 13.9 ± 0.2 31.6 ± 0.9 6.48 ± 0.33 16.3 ± 0.4 50.8 ± 0.4 (E)-2-hexenal −0.922 nd 2.93 ± 0.17 9.81 ± 0.29 1.23 ± 0.01 2.67 ± 0.02 5.96 ± 0.11 (Z)-2-heptenal −0.928 0.799 ± 0.034 6.51 ± 0.23 33.4 ± 0.6 nd 5.12 ± 0.07 20.5 ± 0.5 1-butanol −0.931 0.542 ± 0.033 1.48 ± 0.03 5.15 ± 0.28 0.623 ± 0.006 1.05 ± 0.02 3.87 ± 0.10 3-octen-2-one −0.934 0.465 ± 0.003 5.58 ± 0.16 30.8 ± 0.7 0.755 ± 0.049 5.6 ± 0.9 26.1 ± 0.5 styrene −0.943 1.53 ± 0.02 3.85 ± 0.05 6.64 ± 0.44 1.79 ± 0.05 2.47 ± 0.05 5.99 ± 0.09 1-pentanol −0.946 3.79 ± 0.06 16.8 ± 0.2 66.6 ± 1.5 6.31 ± 0.14 11.7 ± 0.1 49.4 ± 0.5 heptanal −0.976 2.69 ± 0.06 31.5 ± 0.9 187 ± 3 3.25 ± 0.12 22.5 ± 0.3 137 ± 2 aOnly compounds which changed significantly between 0 and 2 months of storage in both dark and light roast almonds are displayed. bCompound identity was confirmed with an authentic standard.

(Acetyloxy)-2-propanone was found in high abundance in fresh Eleven compounds in Table 4 were negatively correlated roasted almonds (Table 4), and similar to 2,3-pentanedione can with liking and increased signiﬁcantly in concentration by two possess a buttery or nutty aroma. The positive aroma attributes months. These compounds were primarily products related to associated with these Maillard products and abundance in fresh lipid oxidation and included hexanoic acid, pentanal, hexanal, 2- almonds supports their positive relationship to consumer pentylfuran, (E)-2-hexenal, (Z)-2-heptenal, 1-butanol, 3-octen- acceptance and the concurrent signiﬁcant decrease in both 2-one, styrene, 1-pentanol, and heptanal.4 Aldehydes typically liking and abundance of these products (Tables 1,2 and 4). possess penetrating aroma characters such as grassy, cucumber,

J DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX ora fArclua n odChemistry Food and Agricultural of Journal ﬁ p μ Table 5. Compounds Found to Have a Signi cant Correlation with Consumer Liking ( < 0.05)a and an Absolute Regression Slope of >2 g/kg Change in Concentration Per Unit Liking with Aroma Characteristics and Concentration in Almonds at 4 and 6 Months

linear R2 slope of conc (μg/kg) vs aroma threshold compound name value liking aroma quality (μg/kg) DR 4 months DR 6 months LR 4 months LR 6 months heptanall,m ,p 0.952 −82.55 fatty, rancid, citrus 50b ,q 85.4 ± 2.0 187 ± 3 87.1 ± 0.9 137 ± 2 1-pentanolm ,p 0.894 −20.87 pungent, fermented, fruity 470c ,q 36.9 ± 0.8 66.6 ± 1.5 34.6 ± 0.3 49.4 ± 0.5 octanall,p 0.885 −93.24 citrus-like, soapy, 55b ,q 69.5 ± 1.5 158 ± 569± 1 127 ± 3 penetrating octanel 0.863 −41.20 gasoline 940d ,q 49.7 ± 1.2 114 ± 1.98 49.7 ± 0.829 69 ± 0.53 nonanall,p 0.856 −72.66 cucumber, lemon peel, 260b ,q 55.5 ± 1.09 117 ± 3.9 52.5 ± 1.48 95 ± 3.02 fatty 1-heptanoll,p 0.852 −14.41 musty, herbal, pungent 425e ,r 9.84 ± 0.301 21.8 ± 0.642 8.73 ± 0.155 17.3 ± 0.251 (E)-2-hexenall,p 0.851 −2.81 penetrating, fatty, green 250b ,q 5.13 ± 0.271 9.81 ± 0.292 5.28 ± 0.0326 5.96 ± 0.114 banana (E)-2-octenall,m ,p 0.846 −20.11 pungent, cucumber, fatty 50−125b ,q 25.1 ± 0.36 46.5 ± 1.19 22.4 ± 0.439 29 ± 0.772 2-heptanonep 0.846 −68.01 fruity, blue cheese, coconut 140e ,r 37.1 ± 0.599 107 ± 2.67 31.9 ± 0.226 78.5 ± 0.807 2-pentylfurann 0.826 −19.62 green bean, metallic, 2000b ,q 28.8 ± 0.583 31.6 ± 0.883 32.4 ± 0.349 50.8 ± 0.429 vegetable γ-hexalactonel 0.815 −13.50 coconut, vanilla, cream 1600f,r 7.66 ± 0.0919 19.3 ± 0.22 6.9 ± 0.128 14.5 ± 0.235 hexanall,m 0.814 −736.91 fatty, green, grassy 75b ,q 1360 ± 15.9 2380 ± 34.6 1390 ± 8.25 1480 ± 6.5 pentanall,m ,p 0.813 −73.57 fermented, bready, fruity 150b ,q 110 ± 1.26 241 ± 10.1 111 ± 0.789 123 ± 0.559 hexanoic acidp ,o 0.810 −95.96 sweaty, cheesy, goaty 700j,q 59.8 ± 2.11 128 ± 4.38 56.3 ± 1.22 106 ± 3.32 heptanel 0.769 −11.10 sweet, ethereal 250 000g ,q 14 ± 0.225 34.2 ± 2.67 15.2 ± 0.669 18.9 ± 0.284 K decanall,p 0.756 −5.72 sweet, citrus, floral 75b ,q 2.44 ± 0.139 6.1 ± 0.367 2.35 ± 0.0381 5.23 ± 0.367 1-octen-3-olm ,p 0.753 −11.27 mushroom, earthy, oily 1e ,r 7.29 ± 0.237 14.9 ± 0.365 5.75 ± 0.158 10 ± 0.207 1-octanoll,p 0.751 −4.26 green, citrus, rose 190e ,r 2.14 ± 0.106 4.7 ± 0.243 1.83 ± 0.0628 3.7 ± 0.131 butanaln ,p 0.746 −7.96 cocoa, musty, malty 25h ,q 8.59 ± 0.131 24.5 ± 1.36 7.89 ± 0.0648 9.23 ± 0.303 pentanoic acid 0.746 −21.60 sickening, putrid, rancid 280i,r 9.43 ± 0.385 23.2 ± 0.574 9.17 ± 0.225 19 ± 0.338 2-octanonep 0.695 −13.34 musty, blue cheese, 190b ,r 4.08 ± 0.127 11.7 ± 0.0845 3.58 ± 0.114 8.96 ± 0.185 mushroom 2-butyl furann ,p 0.662 −6.24 fruity, wine, sweet 10 000g ,q 6.9 ± 0.948 8.7 ± 0.418 8.66 ± 0.138 14.5 ± 0.244 benzaldehydep ,o 0.586 −2.53 artificial almond, sweet, 350k ,r 8.79 ± 0.334 10.6 ± 0.568 9.1 ± 0.0877 8.71 ± 0.195 cherry 2-nonanonep 0.576 −15.57 fruity, soapy, cheese 190b ,r 2.46 ± 0.147 8.56 ± 0.381 2.13 ± 0.0545 5.79 ± 0.263 heptanoic acidp 0.557 −5.96 rancid, cheesy, sweat 100k ,q 1.13 ± 0.101 3.24 ± 0.229 1.12 ± 0.0574 2.51 ± 0.235 octanoic acidp 0.519 −6.68 waxy, rancid, oily 3000j,q 0.863 ± 0.0568 2.77 ± 0.498 0.768 ± 0.289 2.07 ± 0.359 pyrrolep 0.529 5.37 sweet, nutty 20 000e ,r 0.448 ± 0.022 0.243 ± 0.006 0.63 ± 0.0117 0.393 ± 0.027 .Arc odChem. Food Agric. J. 2,3,5- 0.578 2.71 roast, peanut, hazelnut, 90f,r 6.42 ± 0.254 4.93 ± 0.0842 6.03 ± 0.156 5.02 ± 0.0826 trimethylpyrazine 2-methylpropanalp 0.628 20.54 fresh, aldehydic, floral, 3.4f,q 4.49 ± 0.0703 2.94 ± 0.247 4.35 ± 0.143 1.83 ± 0.0769

DOI: furfuralp 0.631 4.63 sweet, almond, bread 3000e ,r 5.2 ± 0.219 3.95 ± 0.12 4.55 ± 0.0394 3.78 ± 0.0424

10.1021/acs.jafc.6b05357 acetoinp 0.652 18.46 sweet, buttery, creamy 800e ,r 4.25 ± 0.105 2.04 ± 0.066 4.17 ± 0.104 2.22 ± 0.0554 XX X,XXX XXX, XXXX, methyl acetatep 0.681 5.96 sweet, fruity, rum nd 4.1 ± 0.107 4.31 ± 0.196 4.3 ± 0.235 2.52 ± 0.237 2-methylpyrazinep 0.731 11.34 chocolate, roasty, nutty 60e ,r 15.8 ± 0.577 12.6 ± 0.479 15.4 ± 0.199 12.1 ± 0.113

2,6-dimethylpyrazine 0.758 2.94 coffee, roasted nuts, cocoa nd 6.09 ± 0.129 4.25 ± 0.0526 6.1 ± 0.0847 4.78 ± 0.148 Article p f,q ± ± ± ±

− 2-methylbutanal 0.781 83.19 fruity, chocolate, nut 10 43.7 0.691 23.1 1.8 47.9 1.17 18.8 0.159 XXX 3-methylbutanalp 0.785 35.66 chocolate, nutty, malty 5.4f,q 19.6 ± 1.09 11.2 ± 0.675 20.3 ± 0.63 8.28 ± 0.141 Journal of Agricultural and Food Chemistry Article

fatty, citrus peel, fruity, and floral (Table 5). Higher alcohols possess fermentative aromas, while 3-octene-2-one contributes 37 0.667 0.0829 a mushroom and earthy character. Organic acids such as ± ± hexanoic acid possess penetrating goaty, sweaty, and cheesy aromas, and 2-pentylfuran has a beany and metallic character. At a certain abundance, all of these compounds would contribute non-natural aroma to almonds and are therefore considered deleterious to almond aroma,5 offering support to the significant negative correlation these compounds have with Bold typeface indicates that

37 liking. 0.718 29.8 0.0339 12.3 ). rmed with an authentic standard. ± ± The significant decrease in average liking of DR and LR fi almonds at two months of storage and concurrent change in volatiles may offer insight into the phenomenon termed “flavor fade” in peanuts. Flavor fade is attributed to the decrease in Recognized autoxidation product of linolenic n positive sensory attributes related to roasted peanut flavor 43 , 35,38,39 39 35 12

, observed early in storage. Reed et al. and Powell et al. 4 ﬂ Recognized autoxidation product of oleic acid methyl attributed avor fade to decreases in concentration from initial l 0.705 36.6 0.297 20.7 . ± ± values of certain pyrazines, as most pyrazines have an inherent 50 nutty or roasted aroma and have been attributed to the main

Ref fl 35 fl k avor of peanuts. Warner et al., however, attributed the avor Compound identity was con . p ff

49 fade of stored peanuts to a masking e ect of increases in lipid www.thegoodscentscompany.com/ 31 ,

12 oxidation aldehydes such as pentanal, hexanal, heptanal, Ref j

. octanal, and nonanal because pyrazine concentrations were 38 48 found not to differ significantly during the storage period. fl Ref Results from our study indicate that avor fade in almonds, as i 1.17 28.8 0.448 11.7 . ± ± indicated by consumer liking, may be a combination of 47 decreases in abundance of compounds associated with fresh 37.1 Ref

h roasted product such as pyrazines and other Maillard products . and concurrent increases in lipid oxidation products. 46 By 6 months of storage, almonds received average hedonic Ref g ratings below 5 (“neither like nor dislike”)(Tables 1 and 2). .

33 Some groups have used hedonic ratings at or below 5 to create r

, an acceptability limit for product acceptance and chemical Ref e f

g/kg) DR 4 months DR 6 months LR 4 months LR 6 months 9,40 .

μ indicators of rancidity. PVs for dark roasted samples and ( 36 1700 light roasted samples at 6 months of storage were 11.36 and aroma threshold Ref

e 2.84 mequiv peroxide/kg oil, respectively. Currently, acceptable . limits for PV are <5 mequiv peroxide/kg oil, and unacceptable 45 PV levels in dark roasted samples would have precluded this

Ref 8

d consumer acceptability limit. In LR almonds, however, PV at 6 .

44 months was only 2.84 and remained below the <5 mequiv

Possible oxidation product from oxidation of 2,4-decadienal. peroxide/kg oil benchmark for the duration of the study. Raisi Ref o c

. et al. also observed PVs below 5 mequiv peroxide/kg oil in 43 , 43 12

, whole, raw almonds with consumer hedonic ratings below 5 on 4 15 beef

Ref a 9-point scale. b Senesi et al. found that acceptance ratings decreased below 5 on a 9-point hedonic scale when PVs ranged from 1.1−1.32 − mequiv O2, and FFA ranged from 4.09 6.70% oleic in whole, g/kg) vs

Recognized autoxidation product of linoleic acid ethyl ester, linoleic acid hydroperoxide, or trilinolein. 11

μ °

m peeled almonds stored under vacuum at 20 C. However, at 43

, the point that hedonic scores were below 5 in DR and LR liking aroma quality 12 ,

4 samples, FFAs were only 0.36 and 0.31% oleic, respectively (Tables 1 and 2). Harris et al. observed similar FFA values in Threshold in aqueous matrix. r

slope of conc ( diced, roasted almonds at the point that consumer hedonic scores fell below 5 on a 9-point scale (0.30% oleic).10 The 2

R results of our study suggest that the current industry limits of PV < 5 mequiv peroxide/kg oil and FFA < 1.5% oleic do not rmed with authentic standards are shown. All aroma descriptors obtained from Good Scents Company ( value 0.7860.860 16.71 17.31 cocoa, roasted nut, roast nd nd 20.7 ﬁ linear correspond with consumer liking (average hedonic score below 5 on a 9-point scale). p Concentration of headspace volatiles has been used to more directly assess perceptual changes in ﬂavor and to supplement 6,9,41 p rancidity indicators such as PV, FFA, and CD. For example, hexanal is frequently detected in tree-nut products subjected to oxidation and is widely considered a good compound name propanol dimethylpyrazine

Only compounds con Threshold in aliphatic/oil matrix. 3,9,41,42 1-chloro- 2- 2,5- acid methyl ester, linolenic acid hydroperoxide, or trilinolenin. Table 5. continued a q compound concentration is aboveester, the oleic published acid threshold. hydroperoxide, or triolein. indicator of oxidation. To examine which headspace

L DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article volatiles may best correlate with consumer liking and therefore Headspace volatiles such as heptanal, octanal, nonanal, and serve as an effective indicator of rancidity in accordance with hexanal displayed good correlation with consumer liking across consumer acceptance, average liking was regressed to relative samples tested and displayed a large degree of concentration concentrations of confirmed headspace compounds (Table 5). change per unit change in hedonic score. Concentration limits Table 5 indicates that many of the measured headspace for these volatiles are not currently established for almond compounds are strongly and significantly correlated with liking samples, and limits in acceptable concentration may be unique (p < 0.05) and are better correlated with liking than PV, CD, to individual processing establishments according to desired and FFAs. Criteria for choosing an optimal indicator may product characteristics and consumer acceptance. Further include: a compound widely detected in almond headspace, a investigation should be done to establish whether these compound for which standards are widely available and indicators correlate well with consumer acceptance under less affordable, and a compound for which there is a large change extreme storage or packaging conditions and across different in concentration per unit change in degree of liking, as this processing variables. would allow for changes in concentration to be detected across a range of method precision and sensitivity. ■ ASSOCIATED CONTENT Using this criteria, heptanal, octanal, nonanal, 2-heptanone, *S Supporting Information hexanal, pentanal, and hexanoic acid, in order of correlation The Supporting Information is available free of charge on the value, are optimal indicators of rancidity. Each of these ACS Publications website at DOI: 10.1021/acs.jafc.6b05357. compounds are widely recognized products of lipid oxidation Includes preliminary data on headspace volatile equili- and display a change of at least 50 μg/kg per unit change in bration, sample calculations of external standardization, liking (Table 5). Furthermore, heptanal, octanal, nonanal, and concentrations of quantitated tentatively identified hexanal, pentanal, and hexanoic acid had similar concentrations fi in both LR and DR almonds at four months of aging, indicating and veri ed compounds in all almond samples for each that these indicators are robust to processing differences during sampling period (PDF) early rancidity development (Table 5). Though heptanal, octanal, and nonanal display a degree of correlation with ■ AUTHOR INFORMATION average consumer liking closer than that of hexanal, hexanal Corresponding Author changed by a much greater concentration per unit change in *Phone: (530) 752-7926; Fax: (530) 752-4759; E-mail: liking (737 μg/kg), making this indicator especially effective in [email protected]. situations where a detection method is less precise and thus less ORCID sensitive to changes in headspace volatile concentration over Alyson E. Mitchell: 0000-0003-0286-5238 time. To identify which volatile compounds negatively correlated Funding fi with consumer liking were responsible for decreases in liking of The Almond Board of California provided nancial support for roasted almonds at four and six months, concentration of this study. volatiles in almond samples were compared to the published Notes sensory threshold (Table 5). Multiple compounds were found The authors declare no competing financial interest. at levels above the sensory threshold in LR and DR almonds stored for four months (heptanal, octanal, hexanal, and 1-octen- ■ ACKNOWLEDGMENTS 3-ol) and DR almonds stored for six months (heptanal, octanal, The authors would like to thank W. Scott Moore and Brian hexanal, 1-octen-3-ol, and pentanal) (Table 5, bold typeface). Dunning of Blue Diamond Almonds for providing almonds and These compounds have aroma characteristics such as rancid, the facilities and expertise to roast almonds. The authors would penetrating, soapy, grassy, fatty, and fungal and are the most fl also like to thank Susan Ebeler and Anna Hjelmeland, UC plausible contributors to undesirable avor changes associated Davis Food Safety and Measurement Facility and Phil Wylie, with oxidative rancidity and associated decreases in liking of Agilent Technologies, for their input on methods development, samples. troubleshooting, and data analysis. The results of this study indicate that certain headspace volatiles correlate better with consumer liking than rancidity ■ ABBREVIATIONS USED indicators such as PV, FFA, and CD (Tables 1−4). For PV and FFA, the recommended industry rejection standard of PV < 5 PV, peroxide value; FFA, free fatty acid value; CD, conjugated ff dienes; LR, light roasted; DR, dark roasted; PTFE, polytetra- mequiv and FFA < 1.5% oleic were not e ective in rejecting our fl samples before the consumer acceptance scores dropped below uoroethylene; MTBE, methyl-tert-butyl-ether; HS-SPME, 5 in the case of FFA for either light or dark roasted samples and headspace solid-phase microextraction; ANOVA, analysis of PVs in light-roasted samples. PV in LR almonds never exceeded variance; MANOVA, multivariate analysis of variance 4 mequiv peroxide/kg oil for the duration of the study, while REFERENCES PV in DR almonds exceeded 16 mequiv peroxide/kg oil. ■ Therefore, PV rejection thresholds should be refined to reflect (1) Yada, S.; Lapsley, K.; Huang, G. A review of composition studies ff ff of cultivated almonds: Macronutrients and micronutrients. J. Food di erences in development of PVs resulting from di erences in − almond processing. In addition, the results of this work and Compos. Anal. 2011, 24, 469 480. (2) Li, S.-C.; Liu, Y.-H.; Liu, J.-F.; Chang, W.-H.; Chen, C.-M.; Chen, other studies indicate that FFAs in heated almond samples C.-Y. O. Almond consumption improved glycemic control and lipid remain at low levels during storage in low humidity (Tables 1 10,11 profiles in patients with type 2 diabetes mellitus. Metab., Clin. Exp. and 2). The rejection threshold of FFA < 1.5% oleic may be 2011, 60, 474−479. best suited for raw almonds exposed to ambient humidity (3) Mexis, S. F.; Kontominas, M. G. Effect of oxygen absorber, rather than roasted almonds. nitrogen flushing, packaging material oxygen transmission rate and

M DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article storage conditions on quality retention of raw whole unpeeled almond (25) Liu, Z.; Lee, H.; Garofalo, F.; Jenkins, D. J. A.; El-sohemy, A. kernels (Prunus dulcis). LWT - Food Sci. Technol. 2010, 43,1−11. Simultaneous Measurement of Three Tocopherols, All-trans-retinol, (4) Frankel, E. N. Lipid oxidation. Prog. Lipid Res. 1980, 19,1−22. and Eight Carotenoids in Human Plasma by Isocratic Liquid (5) Shahidi, F.; John, J. A. Oxidative rancidity in nuts. In Improving Chromatography. J. Chromatogr. Sci. 2011, 49, 221−227. the Safety and Quality of Nuts; Harris, L. J., Ed.; Woodhead Publishing (26) R Development Core Team. R: A language and environment for Limited: Philadelphia, PA, 2013; Vol. 250, pp 198−229. statistical computing. 2003. https://www.r-project.org/ (accessed 11/ (6) Gebreselassie, E.; Clifford, H. Oxidative Stability and Shelf Life of 29/2016). Crackers, Cookies, and Biscuits. In Oxidative Stability and Shelf Life of (27) Lin, X.; Wu, J.; Zhu, R.; Chen, P.; Huang, G.; Li, Y.; Ye, N.; Foods Containing Oils and Fats; Hu, M., Jacobsen, C., Eds.; Academic Huang, B.; Lai, Y.; Zhang, H.; et al. California Almond Shelf Life: Lipid Press and AOCS Press: San Diego, CA, 2016; pp 461−478. Deterioration During Storage. J. Food Sci. 2012, 77, 583−593. (7) García-Pascual, P.; Mateos, M.; Carbonell, V.; Salazar, D. M. (28) Taitano, L. Z.; Singh, R. P. Predictive Modeling of Textural Influence of storage conditions on the quality of shelled and roasted Quality of Almonds During Commercial Storage and Distribution. In almonds. Biosyst. Eng. 2003, 84, 201−209. Advances in Food Process Engineering Research and Applications; (8) Huang, G. Almond Shelf Life Factors; Almond Board of California: Yanniotis, S., Taoukis, P., Stoforos, N., Karathanos, V. T., Eds.; Modesto, CA, 2014. Springer Science+Business Media: New York, 2013; pp 41−59. (9) Mexis, S. F.; Badeka, A. V.; Kontominas, M. G. Quality evaluation (29) Vickers, Z.; Peck, A.; Labuza, T.; Huang, G. Impact of Almond of raw ground almond kernels (Prunus dulcis): Effect of active and Form and Moisture Content on Texture Attributes and Acceptability. modified atmosphere packaging, container oxygen barrier and storage J. Food Sci. 2014, 79, 1399−1406. − conditions. Innovative Food Sci. Emerging Technol. 2009, 10, 580 589. (30) Ling, B.; Hou, L.; Li, R.; Wang, S. Thermal treatment and (10) Harris, N. E.; Westcott, D. E.; Henick, A. S. Rancidity in storage condition effects on walnut paste quality associated with − almonds: shelf life studies. J. Food Sci. 1972, 37, 824 827. enzyme inactivation. LWT - Food Sci. Technol. 2014, 59, 786−793. (11) Senesi, E.; Rizzolo, A.; Sarlo, S. Effect of different packaging (31) Morales, M. T.; Rios, J. J.; Aparicio, R. Changes in the volatile conditions on peeled almond stability. Ital. J. Food Sci. 1991, No. 3, − composition of virgin olive oil during oxidation: flavors and off-flavors. 209 218. J. Agric. Food Chem. 1997, 45, 2666−2673. (12) Frankel, E. N. VOLATILE LIPID OXIDATION PRODUCTS. − (32) Parker, J. K. Thermal generation of aroma. In Flavour Prog. Lipid Res. 1983, 22,1 33. Development, Analysis and Perception in Food and Beverages; Parker, J. (13) White, P. J. Conjugated Diene, Anisidine Value and Carbonyl K., Elmore, S., Methven, L., Eds.; Woodhead Publishing, an imprint of Value Analyses. In Methods to assess quality and stability of oils and fat- Elsevier: Cambridge, U.K., 2015; Vol. 273, pp 151−185. containing foods; Warner, K., Eskin, N. A. M., Eds.; AOCS Press: (33) Belitz, H.-D.; Grosch, W.; Schieberle, P. Aroma Compounds. In Champaign, IL, 1995; pp 159−178. Food Chem., 4th ed.; Springer-Verlag: Berlin, 2009; pp 340−402. (14) Buransompob, A.; Tang, J.; Mao, R.; Swanson, B. G. (34) Kim, M.-O.; Baltes, W. On the Role of 2,3-Dihydro-3,5- RANCIDITY OF WALNUTS AND ALMONDS AFFECTED BY dihydroxy-6-methyl-4(H)-pyran-4-one in the Maillard Reaction. J. SHORT TIME HEAT TREATMENTS FOR INSECT CONTROL. J. Agric. Food Chem. 1996, 44, 282−289. Food Process. Preserv. 2003, 27, 445−464. (35) Powell, J. Evaluation of Initial Flavor Fade in Roasted Peanuts (15) Raisi, M.; Ghorbani, M.; Sadeghi Mahoonak, A.; Kashaninejad, using Sensory Analysis, Gas Chromatography-Olfactometry, Gas M.; Hosseini, H. Effect of storage atmosphere and temperature on the Chromatography-Flame Ionization Detection and Chemosensory oxidative stability of almond kernels during long term storage. J. Stored Techniques. Ph.D. Thesis, Virginia Polytechnic Institute, Blacksburg, Prod. Res. 2015, 62,16−21. VA, 2004. (16) Grosso, N. R.; Resurreccion, A. V. Predicting Consumer ́ ́ ́ Acceptance Ratings of Cracker-coated and Roasted Peanuts from (36) Vazquez-Araujo, L.; Enguix, L.; Verdu, A.; García-García, E.; Descriptive Analysis and Hexanal Measurements. J. Food Sci. 2002, 67, Carbonell-Barrachina, A. A. Investigation of aromatic compounds in − toasted almonds used for the manufacture of turron.́ Eur. Food Res. 1530 1537. − (17) St. Angelo, A. J.; Vercellotti, J.; Jacks, T.; Legendre, M. Lipid Technol. 2008, 227, 243 254. − (37) Luebke, B. The Good Scents Company. http://www. oxidation in foods. Crit. Rev. Food Sci. Nutr. 1996, 36, 175 224. ff (18) Lawless, H. T.; Heymann, H. Sensory Evaluation of Food: thegoodscentscompany.com/prico r.html (accessed Jan 1, 2017). (38) Warner, K. J. H.; Dimick, P. S.; Ziegler, G. R.; Mumma, R. O.; Principles and Practices, 2nd ed.; Springer-Verlag: New York, 2010. ” (19) Lee, J.; Xiao, L.; Zhang, G.; Ebeler, S. E.; Mitchell, A. E. Hollender, R. Flavor-fade and Off-Flavors in Ground Roasted Peanuts Influence of storage on volatile profiles in roasted almonds (prunus As Related to Selected Pyrazines and Aldehydes. J. Food Sci. 1996, 61, − dulcis). J. Agric. Food Chem. 2014, 62, 11236−11245. 469 472. ’ (20) Zhang, G.; Huang, G.; Xiao, L.; Seiber, J.; Mitchell, A. E. (39) Reed, K. A.; Sims, C. A.; Gorbet, D. W.; O Keefe, S. F. Storage Acrylamide formation in almonds (Prunus dulcis): Influences of water activity affects flavor fade in high and normal oleic peanuts. Food − roasting time and temperature, precursors, varietal selection, and Res. Int. 2002, 35, 769 774. ́ storage. J. Agric. Food Chem. 2011, 59, 8225−8232. (40) Gimenez, A.; Ares, F.; Ares, G. Sensory shelf-life estimation: A (21) AOCS Official Method Cd 8-53: Peroxide Value, Acetic Acid- review of current methodological approaches. Food Res. Int. 2012, 49, Chloroform Method. In Official Methods and Recommended Practices of 311−325. the American Oil Chemists’ Society; Firestone, D., Ed.; AOCS Press: (41) Pastorelli, S.; Valzacchi, S.; Rodriguez, A.; Simoneau, C. Solid- Champaign, IL, 1997. phase microextraction method for the determination of hexanal in (22) AOCS. Method Cd 3d-63 Acid Value. In Official Methods and hazelnuts as an indicator of the interaction of active packaging Recommended Practices of the American Oil Chemists’ Society; Firestone, materials with food aroma compounds. Food Addit. Contam. 2006, 23, D., Ed.; AOCS Press: Champaign, IL, 1997. 1236−1241. (23) Xiao, L.; Lee, J.; Zhang, G.; Ebeler, S. E.; Wickramasinghe, N.; (42) Jensen, P. N.; Danielsen, B.; Bertelsen, G.; Skibsted, L. H.; Seiber, J.; Mitchell, A. E. HS-SPME GC/MS characterization of Andersen, M. L. Storage stabilities of pork scratchings, peanuts, volatiles in raw and dry-roasted almonds (Prunus dulcis). Food Chem. oatmeal and muesli: Comparison of ESR spectroscopy, headspace-GC 2014, 151,31−39. and sensory evaluation for detection of oxidation in dry foods. Food (24) Puspitasari-Nienaber, N. L.; Ferruzzi, M. G.; Schwartz, S. J. Chem. 2005, 91,25−38. Simultaneous detection of tocopherols, carotenoids, and chlorophylls (43) Belitz, H.-D.; Grosch, W.; Schieberle, P. Lipids. Food Chem. in vegetable oils by direct injection C30 RP-HPLC with coulometric 2009, 158−247. electrochemical array detection. J. Am. Oil Chem. Soc. 2002, 79, 633− (44) Aparicio, R.; Luna, G. Characterisation of monovarietal virgin 640. olive oils. Eur. J. Lipid Sci. Technol. 2002, 104, 614−627.

N DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX Journal of Agricultural and Food Chemistry Article

(45) Luna, G.; Morales, M. T.; Aparicio, R. Changes Induced by UV Radiation during Virgin Olive Oil Storage. J. Agric. Food Chem. 2006, 54, 4790−4794. (46) Evans, C. D.; Moser, H. A.; List, G. R. Odor and flavor responses to additives in edible oils. J. Am. Oil Chem. Soc. 1971, 48, 495−498. (47) Przybylski, R.; Eskin, N. A. M. Methods to measure volatile compounds and the flavor significance of volatile compounds. Methods to assess quality and stability of oils and fat-containing foods 1995, 107− 133. (48) Oonbumrung, S. B.; Amura, H. T.; Ookdasanit, J. M.; Akamoto, H. N.; Shihara, M. I. Characteristic Aroma Components of the Volatile Oil of Yellow Keaw Mango Fruits Determined by Limited Odor Unit Method. Food Sci. Technol. Res. 2001, 7, 200−206. (49) Morales, M. T.; Luna, G.; Aparicio, R. Comparative study of virgin olive oil sensory defects. Food Chem. 2005, 91, 293−301. (50) Buttery, R. G.; Stern, D. J.; Ling, L. C. Studies on Flavor Volatiles of Some Sweet Corn Products. J. Agric. Food Chem. 1994, 42, 791−795.

O DOI: 10.1021/acs.jafc.6b05357 J. Agric. Food Chem. XXXX, XXX, XXX−XXX